Deep-learning-based Automated Measurement Of Outer Retinal Layer Thickness for Use in the Assessment of Age-related Macular Degeneration, Applicable to Swept-Source and Spectral-Domain OCT Imaging

Abstract

Purpose : This study aims to establish a deep learning-based automated algorithm, applicable to both swept-source optical coherence tomography (SS-OCT) and spectral-domain OCT (SD-OCT) angiography scans, for measuring outer retinal layer (ORL) thickness and then explore whether ORL thickness can predict the progression of age-related macular degeneration (AMD).

Methods : The algorithm for measuring ORL thickness was developed based on a modified TransUNet model with clinical retinal features manifested in the progression of AMD. The performance of the algorithm was assessed using the intersection over union (IoU) metric. Correlation analysis was employed to compare ORL thickness measurements between SS-OCT and SD-OCT scans. Agreement between these measurements was further evaluated using Bland-Altman analysis. A comprehensive analysis of ORL thickness was studied among various stages of AMD, encompassing 80 normal eyes without any ocular disease, 30 eyes with intermediate AMD (iAMD) with macular reticular pseudodrusen, 49 eyes with iAMD displaying typical soft drusen only, and 40 eyes with late nonexudative AMD characterized by the presence of persistent choroidal hypertransmission defects (hyperTDs).

Results : The algorithm demonstrates a high accuracy with an IoU of 0.9698 in the testing dataset when segmenting the ORL using both SS-OCT and SD-OCT datasets. The robustness and applicability of the algorithm are indicated by strong correlations (r = 0.9551, p < 0.0001 in the fovea centered 3mm circle, and r = 0.9442, p < 0.0001 in the 5mm circle) and agreements (mean bias = 0.5440μm in the 3mm circle, and 1.392μm in the 5mm circle) between the ORL thickness measurements from the SS-OCT and SD-OCT scans. Comparative analysis reveals significant differences (p < 0.0001) in ORL thickness among different AMD stages.

Conclusions : A deep learning-based algorithm was developed to measure ORL thickness as a surrogate marker for assessing AMD severity. The comparative analysis revealed significant differences in ORL thickness measurements among different AMD stages, indicating its potential use as an independent biomarker for predicting AMD progression.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.